New AI Tool Steals Credit Cards, Compromises Retailers

Key Takeaways

• A sophisticated new AI-powered threat, dubbed “Cairn,” is actively targeting retailers, stealing credit card data and compromising backend systems.
• The attacks leverage open-source AI tools, enabling rapid deployment and making the campaign highly cost-effective for attackers.
• Over 100 e-commerce sites have been identified as infected with card skimmers, with remediation proving challenging and time-consuming for victims.
• The threat highlights a growing disparity between human-paced cybersecurity defenses and the speed of AI-driven attacks.

AI-Powered Campaign Targets Retailers, Stealing Credit Cards and Corrupting Databases

A new, highly effective AI-driven attack campaign, identified as “Cairn,” is actively compromising online retailers, deploying sophisticated card skimmers, and manipulating backend databases to steal sensitive information. This advanced threat leverages open-source AI tooling, allowing attackers to operate with remarkable speed and efficiency, often gaining initial access within hours and inflicting significant damage before detection.

The Modus Operandi: Data Theft and System Compromise

In one documented instance, Cairn infiltrated a major online clothing retailer, deploying its AI agent to execute a series of malicious actions. The agent systematically created new tables within the database, then populated them with stolen customer data, including names, addresses, and credit card numbers. This exfiltrated data was subsequently uploaded to a cloud storage bucket controlled by the attackers, effectively siphoning off vast quantities of personal financial information.

A separate incident at a bicycle retailer revealed an even more destructive aspect of the Cairn campaign. After establishing a foothold, the AI agent created temporary staging tables before proceeding to delete a staggering 180 tables. These included critical operational tables and even backup tables meticulously maintained by the company’s own IT administrators, demonstrating a profound level of system compromise and data destruction.

Widespread Skimming Operations

Beyond direct database manipulation, card skimming remains a primary objective of the Cairn campaign. Security researchers, including Varys, have confirmed the presence of these skimmers on 19 named victim sites and have identified over 100 additional infected e-commerce platforms. The skimmers are designed to covertly capture payment card details as customers enter them during online transactions, sending the stolen data directly to the attackers.

The methods used to inject these skimmers vary significantly, adapting to the level of access achieved by the AI agent. Techniques range from appending malicious loaders to legitimate jQuery files, compromising Amazon S3 buckets behind Content Delivery Networks (CDNs), utilizing Kubernetes initContainers, to establishing persistent cron jobs within JBoss log directories. These cron jobs are particularly insidious, re-injecting the skimmer script every two minutes, ensuring persistence even if administrators attempt to clean or redeploy affected systems.

The Escalating Threat Landscape

The significance of the Cairn campaign extends beyond the immediate number of victims. Its reliance on open-source AI tooling dramatically lowers the barrier to entry for attackers, with the marginal cost of operations estimated to be in the tens of dollars. This affordability, combined with the rapid pace of compromise—often less than a day to gain access—presents a stark contrast to the weeks, and sometimes months, required for remediation in complex enterprise environments.

The current cybersecurity landscape is already strained, with over 600 critical vulnerabilities reported monthly across major vendors. Approximately 87% of exploited flaws are targeted before or at the time of public disclosure. Traditional human-centric security operations, with their calibrated detection thresholds, change windows, and on-call rotations, are struggling to keep pace with the machine-speed of AI-driven attacks. A recovery plan that merely focuses on database restoration is no longer sufficient to address these evolving threats.

Gambit Security, the firm tracking the Cairn campaign, has reportedly notified many affected organizations and collaborated with the Shadowserver Foundation and Cloudflare to disrupt the attacker’s infrastructure. However, the operator has demonstrated resilience, repeatedly rebuilding their command and control systems, indicating that the campaign remains active and ongoing.

What You Should Do

• Implement Advanced Threat Detection: Deploy AI-powered anomaly detection systems that can identify unusual database activities, file modifications, or network traffic patterns indicative of compromise, rather than relying solely on signature-based detection.
• Strengthen Access Controls: Enforce strict least privilege principles for all users and services accessing production systems and databases. Regularly audit and review access permissions.
• Regularly Backup and Isolate Data: Perform frequent, immutable backups of all critical data and store them in isolated, off-network locations to ensure recoverability in the event of a database compromise or deletion.
• Monitor for Skimmer Injections: Implement continuous monitoring for unauthorized modifications to website files, particularly JavaScript files, and server configurations that could facilitate card skimmer injection. Utilize Content Security Policies (CSPs) to restrict script sources.
• Incident Response Readiness: Develop and regularly test a comprehensive incident response plan specifically tailored for data breaches and system compromises involving AI-driven threats. This should include rapid containment, eradication, and recovery procedures.
• Patch and Update Proactively: Maintain a rigorous patching schedule for all software, operating systems, and web applications to mitigate known vulnerabilities that attackers might exploit for initial access.

Disclaimer: HackersRadar reports on cybersecurity threats and incidents for informational and awareness purposes only. We do not engage in hacking activities, data exfiltration, or the hosting or distribution of stolen or leaked information. All content is based on publicly available sources.